Found a cool grant for solo founders, up to $1000 from Merge Club. I’m applying, and thought some of you might want to check it out too:
https://merge.club/noemie
https://merge.club/noemie
merge.club
Noémie - Grantsmith at Merge Club
Noémie joins Merge as a Grantsmith supporting ambitious technology projects.
Vercel and Convex are experiencing a massive outage
Almost all webapps that use convex or vercel are down
Almost all webapps that use convex or vercel are down
will to code
Vercel and Convex are experiencing a massive outage Almost all webapps that use convex or vercel are down
Its caused by an AWS outage
A lot of services are impacted
Except Cloudflare
A lot of services are impacted
Except Cloudflare
image_2025-10-21_21-51-47.png
168.6 KB
I just finished integrating the agent
It has low agency now but I plan to do a deep dive on prompt engineering and make it as good as v0. That's the benchmark.
It's all about being patient and writing text. How hard could it be?
#buildinpublic
It has low agency now but I plan to do a deep dive on prompt engineering and make it as good as v0. That's the benchmark.
It's all about being patient and writing text. How hard could it be?
#buildinpublic
🔥4
I wanted to create an X for tech people (devs, designers, investors, startups...) it would be such a cool platform. but then I immediately gave up thinking I needed a lot of capital for hosting images and videos... But now after looking at cloudflare's pricing... Its not impossible.
The concept of "social media" has grown stale from lack of innovation. Primarily because its already making money. This needs to change. We should create our own platform that incentives our own values.
The Algorithm.
- Will be open source
- Will be a program you can run on your computer
- Imagine being able to test and refine the algorithm just like you mine Bitcoin. First you download a chunk of the data which represents posts, then you tweak some parameters and see what kind of posts the algorithm favors... then share it... it gets merged. This should be the future!
Control the Algorithm, Control Fate.
An algorithm is so powerful that it rewires people's brains. We need to control it. Seize the means of attention!
X has to serve different kinds of people so it will never be perfect, it will always prioritize money. We must break away from its hold and build our own algorithm which we control.
In this day and age tribe means everything. The algorithms of today profit by dividing people into tribes. My goal is to create a single tribe of tech people that will create invest and build the future together. Let us take the algorithms onto our own hands!! Programmers UNITE.
Also another cool concept would be to wipe all content after 3 months... a clever way to save money 😅 and an interesting game theory to introduce to the system
The concept of "social media" has grown stale from lack of innovation. Primarily because its already making money. This needs to change. We should create our own platform that incentives our own values.
The Algorithm.
- Will be open source
- Will be a program you can run on your computer
- Imagine being able to test and refine the algorithm just like you mine Bitcoin. First you download a chunk of the data which represents posts, then you tweak some parameters and see what kind of posts the algorithm favors... then share it... it gets merged. This should be the future!
Control the Algorithm, Control Fate.
An algorithm is so powerful that it rewires people's brains. We need to control it. Seize the means of attention!
X has to serve different kinds of people so it will never be perfect, it will always prioritize money. We must break away from its hold and build our own algorithm which we control.
In this day and age tribe means everything. The algorithms of today profit by dividing people into tribes. My goal is to create a single tribe of tech people that will create invest and build the future together. Let us take the algorithms onto our own hands!! Programmers UNITE.
Also another cool concept would be to wipe all content after 3 months... a clever way to save money 😅 and an interesting game theory to introduce to the system
🔥5
will to code
v0 systems prompts leaked and its 2000 lines long. Yeah, Prompt Engineering is a real job just like that, a couple text files leak and they lost their edge. The repo already got 3.4k stars and 2k forks
lmao create the problem then sell the solution
The guy that leaked all the sys prompts is charging the companies for his service
Also his repo is at 93K stars. It was at 3.4K stars when I shared it
The guy that leaked all the sys prompts is charging the companies for his service
Also his repo is at 93K stars. It was at 3.4K stars when I shared it
🔥1
Its #friday. Let us relax, chill and build something cool.
Project for today. A simple bot that will let users generate images and videos in group chats. Then convert them into stickers and Gifs.
Project for today. A simple bot that will let users generate images and videos in group chats. Then convert them into stickers and Gifs.
🔥3
here's the bot @the_image_gen_bot
Unfortunately no sora 2
I spent a lot of time trying to make it work with deno but it just didn't. But it's fine, I'll build it with a different stack next friday
Unfortunately no sora 2
I spent a lot of time trying to make it work with deno but it just didn't. But it's fine, I'll build it with a different stack next friday
🔥5
will to code
Talk is cheap, build it from scratch
While coding I can't help but think about what Karpathy said.
When it comes time to write code, all your abstractions fall apart. Talk is a cheap token. It never converts well into action. The most expensive token is experience
Always Always Always get your hands dirty in the nitty gritty
Only by building it from scratch can you understand it
And the funny thing is that when your abstractions collapse which they always do you just throw your hands in the air and say "its too complicated", "i tried"
The Joy of suffering >> >>
The Joy of acquiring expensive tokens and saying yes I earned this with my own inference
When it comes time to write code, all your abstractions fall apart. Talk is a cheap token. It never converts well into action. The most expensive token is experience
Always Always Always get your hands dirty in the nitty gritty
Only by building it from scratch can you understand it
And the funny thing is that when your abstractions collapse which they always do you just throw your hands in the air and say "its too complicated", "i tried"
The Joy of suffering >> >>
The Joy of acquiring expensive tokens and saying yes I earned this with my own inference
⚡4
Right now most of AI Video is brainrot
However, as soon as the tools mature, it will be pure entertainment (higher quality than tiktok could ever dream of)
Think endless funny and amazing fan made spin offs or scenarios involving your favorite tv shows, authors, philosophers and characters
It's coming
However, as soon as the tools mature, it will be pure entertainment (higher quality than tiktok could ever dream of)
Think endless funny and amazing fan made spin offs or scenarios involving your favorite tv shows, authors, philosophers and characters
It's coming
will to code
here's the bot @the_image_gen_bot Unfortunately no sora 2 I spent a lot of time trying to make it work with deno but it just didn't. But it's fine, I'll build it with a different stack next friday
Working on this literally validated my current project. Most of the time 99% (4+ hours) was wasted on getting used to the new deno console plus the model apis quirks....
My new app eliminates all of this, had I been using my own service I would've finished the bot in <45 mins with sora 2 support and many other generative media features
That's why this week I'll try to make the API work and on Friday I'll build the bot on my own infra!
My new app eliminates all of this, had I been using my own service I would've finished the bot in <45 mins with sora 2 support and many other generative media features
That's why this week I'll try to make the API work and on Friday I'll build the bot on my own infra!
🔥3
will to code
Working on this literally validated my current project. Most of the time 99% (4+ hours) was wasted on getting used to the new deno console plus the model apis quirks.... My new app eliminates all of this, had I been using my own service I would've finished…
Timeline so far
- Monday: Built the API layer
- Tuesday: Switch to Cloudflare for Videos and Images
- Today: Added Support for Video Generation Models
#buildinpublic
- Monday: Built the API layer
- Tuesday: Switch to Cloudflare for Videos and Images
- Today: Added Support for Video Generation Models
#buildinpublic
🔥6
will to code
An algorithm is so powerful that it rewires people's brains. We need to control it. Seize the means of attention!
We could do it better and it wouldn't even be that difficult
X's incentive is to monetize everyone, ours will be different
X's incentive is to monetize everyone, ours will be different